Louisiana State University LSU Digital Commons
LSU Master's Theses Graduate School
2013 Assessing the efficiency of alternative best management practices to reduce nonpoint source pollution in the broiler production region of Louisiana Bryan Gottshall Louisiana State University and Agricultural and Mechanical College
Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_theses Part of the Agricultural Economics Commons
Recommended Citation Gottshall, Bryan, "Assessing the efficiency of alternative best management practices to reduce nonpoint source pollution in the broiler production region of Louisiana" (2013). LSU Master's Theses. 1783. https://digitalcommons.lsu.edu/gradschool_theses/1783
This Thesis is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Master's Theses by an authorized graduate school editor of LSU Digital Commons. For more information, please contact [email protected].
ASSESSING THE EFFICIENCY OF ALTERNATIVE BEST MANAGEMENT PRACTICES TO REDUCE NONPOINT SOURCE POLLUTION IN THE BROILER PRODUCTION REGION OF LOUISIANA
A Thesis
Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Master of Science
in
The Department of Agricultural Economics and Agribusiness
by Bryan Gottshall B.S., Mercy College, 2009 December 2013
Acknowledgements
I would first and foremost like to thank my parents Kathleen McLeod and Thomas
Gottshall for their love and support throughout my life. Without the support of my family, I would never have made it anywhere in life, much less through graduate school.
I would like to thank my major advisor Dr. Krishna Paudel for providing the opportunity to conduct this study as well as constant guidance and encouragement throughout. I would also like to thank my committee, Dr. Jeffery Gillespie and Dr. John Westra for their suggestions and guidance as well as Niu Huizhen for her help collecting data and creating the GIS layers.
I would like to express gratitude for the constant help and support from my fellow students at LSU. In particular I’d like to thank Basu Bhandari, Abhishek Bharad, Luke Boutwell,
Jill Christoferson, Madhav Regmi and Franklin Vaca for their constant academic support and therapeutic counsel. I am deeply indebted to Mahesh Pandit for his unending technical support.
Finally, I would like to thank all of my friends who encouraged me to stick with this program despite my constant threats to quit. In particular I would like to thank Ian Smith, Zach
Wilkerson and Roxanne Guidry for their unfettering friendship throughout this period. I would also like to thank my girlfriend, Missy Wilkinson, for her unending love, support and encouragement as well as her help in editing this thesis.
ii
Table of Contents Acknowledgements ...... ii
List of Tables ...... v
List of Figures ...... vii
List of Abbreviations ...... viii
Abstract ...... x
Chapter 1. Introduction ...... 1 1.1 Background ...... 1 1.2 Problem Statement ...... 4 1.3 Objectives ...... 5 1.4 Overview of Thesis ...... 5
Chapter 2. Literature Review ...... 6 2.1 Water Policy: A brief history ...... 6 2.2 BMP Efficiency ...... 14 2.3 Optimization and GIS in Determining BMP Cost Effectiveness...... 21
Chapter 3. Data and Methodology ...... 27 3.1 Data and Study Area ...... 27 3.2 Watershed Modeling ...... 35 3.3 BMP Reduction Coefficients and Optimization Programming ...... 39
Chapter 4. Results ...... 43 4.1 Chenerie Creek...... 43 4.2 Bayou Desiard ...... 50 4.3 Wet and Dry Years in Chenerie Creek ...... 57 4.4 Wet and Dry Years in Bayou Desiard...... 68 4.7 Model Self Calibration ...... 79
Chapter 5. Summary and Conclusions ...... 80 5.1 Summary ...... 80 5.2 Policy Implications ...... 82 5.3 Further Research and Study Limitations...... 86
References ...... 89
Appendix A Regression Summary...... 97
Appendix B: Optimization Matrix Examples and Cost per Hectare Reduction Figures ...... 102
iii
Appendix C GIS Layers ...... 108
Vita ...... 113
iv
List of Tables
Table 3.1 BMP cost per land unit……………………………………………………………...... 35
Table 4.1 Summary of Total Pollutant Reduction Costs at Different Levels of Targeted Phosphorus Reduction for Chenerie Creek…………………………………………...45
Table 4.2 Summary of BMPs Adopted and Land Use in Chenerie Creek.………………...……46
Table 4.3 Shadow Prices of Constraints in Chenerie Creek……………………………………..48
Table 4.4 Reduced Costs of BMPs in Chenerie Creek…………………………………………..49
Table 4.5 Summary of Total Pollutant Reduction Costs at Different Levels of Targeted Phosphorus Reduction for Bayou Desiard……………………………………………51
Table 4.6 Summary of BMPs Adopted and Land Use in Bayou Desiard……………………….52
Table 4.7 Shadow Prices of Constraints in Bayou Desiard…………………………………...... 54
Table 4.8 Reduced Costs of BMPs in Bayou Desiard………………………………………...... 56
Table 4.9 Summary of Total Pollutant Reduction Costs at Different Levels of Targeted Phosphorus Reduction for Chenerie Creek in a Wet Year (2004) ...... 59
Table 4.10 Summary of BMPs Adopted and Land Use in Chenerie Creek in a Wet Year (2004)………………………………………………………………………………...60
Table 4.11 Shadow Prices of Constraints for Chenerie Creek in a Wet Year (2004)……………61
Table 4.12 Reduced Costs of BMPs for Chenerie Creek in a Wet Year (2004)………………....62
Table 4.13 Summary of Total Pollutant Reduction Costs at Different Levels of Targeted Phosphorus Reduction for Chenerie Creek in a Dry Year (2010)…………………...64
Table 4.14 Summary of BMPs Adopted and Landuse for Chenerie Creek in a Dry Year (2010)……………………………………………………………………………...... 65
Table 4.15 Shadow Prices of Constraints for Chenerie Creek in a Dry Year (2010)……………66
Table 4.16 Reduced Costs of BMPs for Chenerie Creek in a Dry Year (2010)…………………67
Table 4.17 Summary of Total Pollutant Reduction Costs at Different Levels of Targeted Phosphorus Reduction Bayou Desiard in a Wet Year (2004)………………………..69 v
Table 4.18 Summary of BMPs Adopted and Land Use for Bayou Desiard in a Wet Year (2004)………………………………………………………………………………...70
Table 4.19 Shadow Prices of Constraints for Bayou Desiard in a Wet Year (2004)………….....72
Table 4.20 Reduced Costs of BMPs for Bayou Desiard in a Wet Year (2004)……………….....73
Table 4.21 Summary of Total Pollutant Reduction Costs at Different Levels of Targeted……...75 Phosphorus Reduction for Bayou Desiard in a Dry Year (2005)
Table 4.22 Summary of BMPs Adopted and Land Use for Bayou Desiard in a Dry Year (2005)………………………………………………………………………………...76
Table 4.23 Shadow Prices of Constraints for Bayou Desiard in a Dry Year (2005)…………….77
Table 4.24 Reduced Costs of BMPs for Bayou Desiard in a Dry Year (2005)………………….78
Table A.1 Sample Data Table for Cover Crops in Chenerie Creek……………………………...97
Table A.2 Regression Coefficients for Chenerie Creek………………………………………….99
Table A.3 Regression Coefficients for Bayou Desiard…………………………………………100
Table A.4 Regression Coefficients for Chenerie Creek in Wet and Dry Years .………………..100
Table A.5 Regression Coefficients for Bayou Desiard in Wet and Dry Years…………………101
Table B.1 Per Hectare Unit Reduction Costs for BMPs in Chenerie Creek .……...... 105
Table B.2 Per Hectare Unit Reduction Costs for BMPs in Bayou Desiard…………………….105
Table B.3 Per Hectare Unit Reduction Costs for BMPs in Chenerie Creek for a Wet Year…...106
Table B.4 Per Hectare Unit Reduction Costs for BMPs in Chenerie Creek for a Dry Year .…...106
Table B.5 Per Hectare Unit Reduction Costs for BMPs in Bayou Desiard for a Wet Year……107
Table B.6 Per Hectare Unit Reduction Costs for BMPs in Bayou Desiard for a Dry Year .……107
vi
List of Figures
Figure 3.1 Louisiana Broiler Production Region………………………………………………...27
Figure 3.2 Map of Cheneire Creek Watershed…………………………………………………..29
Figure 3.3 Map of the Bayou Desiard Watershed……………………………………………...... 30
Figure 3.4 Optimization Procedure……………………………………………………………....41
Figure 5.1 Unit Cost of Phosphorus Reduction………………………………………………….83
Figure A.1 Regression Sample for Cover Crops in Chenerie Creek…………………………….99
Figure B.1 Optimization Matrix for Chenerie Creek at 10% Phosphorus Reduction .………….102
Figure B.2 Optimization Matrix for Chenerie Creek at 10% Phosphorus Reduction in a Wet Year……………………………………………………………………………103
Figure B.3 Optimization Matrix for Chenerie Creek at 10% Phosphorus Reduction in a Dry Year……...... 104
Figure C.1 AFO Layers…………………………………………………………………………108
Figure C.2 Basin Layer………………………………………………………………………....108
Figure C.3 County Layer……………………………………………………………………….109
Figure C.4 DEM Layer…………………………………………………………………………109
Figure C.5 Land Use Layer……………………………………………………………………..110
Figure C.6 Physiographic Province Layer……………………………………………………...110
Figure C.7 River Layer………………………………………………………………………....111
Figure C.8 Soil Layer…………………………………………………………………………...111
Figure C.9 Soil Test P Layer…………………………………………………………………...112
Figure C.10 Weather Layer…….……………………………………………………………….112
vii
List of Abbreviations
AEU Animal Equivalent Unit
AFO Animal Feeding Operation
AVGWLF ArcView Generalized Watershed Loading Function
BMP Best Management Practices
CAC Command and Control
CAFO Concentrated Animal Feeding Operation
CRP Conservation Reserve Program
CWA Clean Water Act
EQIP Environmental Quality Incentive Program
EPA Environmental Protection Agency
FBCP Farm Bill Conservation Program
FWP Fish and Wildlife Propagation
GA Genetic Algorithm
GIS Geographic Information System
GWLF-E Generalized Watershed Loading Function Enhanced
HSPF Hydraulic Simulation Program - Fortran k Soil Erosion Factor
LBPR Louisiana Broiler Production Region
LDEQ Louisiana Department of Environmental Quality
L-S Slope Length Factor
NOAA National Ocean and Atmospheric Administration
viii
NPS Nonpoint Source
NRCS National Resource Conservation Service
NPDES National Pollution Discharge Elimination System
PCR Primary Contact Recreation
PRediCT Pollution Reduction Comparison Tool
PS Point Source
RCWP Rural Clean Water Program
RHS Right Hand Side
SCR Secondary Contact Recreation
SWAT Soil and Water Assessment Tool
SWMM Storm Water Management Model
TMDL Total Maximum Daily Load
USDA United States Department of Agriculture
USLE Universal Soil Loss Equation
WQA Water Quality Act
WQMP Louisiana Water Quality Management Plan
ix
Abstract
The Louisiana broiler production region is located in North Central and Northwestern
Louisiana. The region consists of twelve parishes in Northwestern and North Central Louisiana.
The broiler production region is a significant contributor of nonpoint source (NPS) pollution to nearby waterways. This pollution is a consequence of sediment, nitrogen and phosphorus runoff caused by agricultural production. NPS pollution is difficult to mitigate due to uncertainties in its point of origin as well as a host of other factors ranging from rainfall to topographical parameters. Best Management Practices (BMPs) have been shown to be a reliable method for reducing nonpoint source pollution emanating from agricultural production. To reduce pollutants, several BMPs have been recommended, specific to crops and regions, by the Natural
Resource Conservation Service of the United States Department of Agriculture (NRCS/USDA) and U.S. Environmental Protection Agency (EPA). Successful implementation of BMPs for water quality improvement requires careful study of both nonpoint pollution sources and their effectiveness in a given spatial situation. These assessments are being conducted for several watersheds throughout the United States; however, many watersheds in Louisiana remain unexamined. This study focuses on two watersheds in the broiler production region of Louisiana and utilizes a GIS based simulation program to determine the best least cost solution for the application of BMPs in the study region. Analyses were conducted under alternative climate change and BMP effectiveness scenarios. Results indicate that it is cost effective to implement nutrient management to reduce phosphorus pollution.
x
Chapter 1. Introduction
1.1 Background
Water quality has become a major concern in the United States and throughout the world. The Clean Water Act (CWA) of 1972 laid the groundwork for several programs to improve water quality; however, according to the Environmental Protection
Agency, over 40% of assessed waterways do not achieve minimum requirements for their intended use (USEPA, 2008).
Under the National Pollution Discharge Elimination System (NPDES) program, all point source (PS) polluters, such as municipal and industrial waste facilities, are required to obtain permits, which are managed by the EPA and state environmental agencies. The CWA requires states to determine a Total Maximum Daily Load (TMDL) for each watershed that does not meet its intended use. The NPDES program is widely recognized as effectively reducing point source pollution and restoring waterways to their designated uses throughout the United States. However, due to their diffuse nature, nonpoint source polluters, which are exemplified by emissions from mobile sources, leaching or runoff from agricultural lands and runoff from residential areas and construction sites, remain an unresolved cause of water quality problems, despite the determination of TMDLs (USEPA, 2003). Agricultural runoff has been found to be the largest single contributor to nonpoint source pollution in rural watersheds, contributing an estimated 65% of the nitrogen pollution to the Gulf of Mexico (USEPA, 2000).
1
Agriculture is the major source of nonpoint pollution in northern Louisiana. While fertilizers have become necessary for modern agricultural production, they can leach nutrients, such as nitrogen and phosphorus, into surrounding waterways when applied improperly. Nutrient runoff results in eutrophication when it reaches nearby waterways, which often leads to hypoxia. These hypoxic zones are a growing problem in the Gulf of
Mexico, threatening gulf ecosystems and the many nearby industries dependent on marine life (Hall, 2009).
A diverse range of structural and management methods known as best management practices (BMPs) are widely used to control NPS runoff. Implementation of
BMPs at critical locations throughout various watersheds have been shown to improve water quality in waterways compromised by NPS (Zhen et al., 2004). However, BMPs are being implemented without sufficient studies at the farm or watershed level to determine which combination of BMPs is most effective (NRCS, 2004).
When determining which BMPs to utilize in a region, cost and farmers’ willingness to adopt these practices are important factors to consider. Farmers or agricultural producers who are not likely to adopt expensive BMPs often absorb implementation costs. Moreover, farm managers are interested in maximizing profits while societal and environmental institutions are focused on improving water quality.
These public groups are not likely to be concerned with pollution reduction costs unless implementation of these practices causes them economic burden. However, achieving the least cost solution is often in concert with both public and private interests (Gitau et al.,
2004).
2
The key challenge of structuring an effective BMP program targeted at reducing
NPS pollution is achieving a maximum reduction in pollutant loading at a minimum cost
(Giri et al., 2012). Selection and placement of BMPs have been shown to be nearly three times more cost effective than targeting methods of specified pollutants (Arabi et al.,
2006). Once BMPs have been implemented, pollutant reduction from that site can be satisfactorily measured over time. However, predetermining the impact of BMPs on a specific site is generally more complicated (Gitau et al., 2004).
Over the past several years, a multitude of models and simulation programs have been developed to predict the best combination of BMPs in a given watershed. Modeling techniques include linear programming, Monte Carlo simulation, scatter search and sorted genetic algorithms. Geographic information system (GIS) based simulation software includes the Soil and Water Assessment Tool (SWAT), Mapshed, the Pollution
Reduction Comparison Tool (PRedICT), the Storm Water Management Model (SWMM) and the Hydraulic Simulation Program - Fortran (HSPF). Arabi et al. (2006) used SWAT and a genetic algorithm (GA) to optimize BMPs for watersheds in Indiana. Mishra et al.
(2007) employed SWAT to identify areas of high sediment yield and determine structural designs to minimize them. Kaini et al. (2012) utilized a single objective optimal control model in concert with SWAT and a GA to determine optimal BMP placement for the
Silver Creek watershed in Illinois.
In this study, we utilize Mapshed simulation software in conjunction with an optimization model to determine a cost-effective BMP combination for the Chenerie
Creek and Bayou Desiard, two watersheds within the broiler production region. Multiple
3
studies have utilized Mapshed, or its predecessor the ArcView Generalized Watershed
Loading Function (AVGWLF), in combination with various optimization techniques to determine optimal BMP placement (McGarety el al., 2005; Markel et al., 2007; Georgas et al., 2009). This software package is a useful tool which can be utilized to reduce nutrient (nitrogen and phosphorus) and sediment loads in waterways and help them reach their predetermined TMDL goals.
1.2 Problem Statement
Recent literature indicates concern over the inability of previous pollution reduction efforts to identify and reduce NPS polluters. While PS pollution has been largely reduced in the United States, NPS remains a challenge to environmental regulators, as it is difficult to identify and costly to monitor sources due to their diffuse nature. Further problems include challenges introduced by topographical, hydrological and climatic factors and their influence over the flow of pollutants. Agricultural runoff has been identified as a major contributor to NPS pollution. BMPs have been shown to reduce effluent runoff from agricultural production. However, the factors that dictate the optimal BMPs to utilize in a specific region vary drastically over a small area, making precise measurement at the watershed and sub-watershed levels necessary for ideal implementation. This research utilizes simulation software to determine the effectiveness of several BMPs in the Chenerie Creek and Bayou Desiard watersheds, two watersheds located in the Louisiana Broiler Production Region (LBPR). The results of this analysis are analyzed, along with implementation cost information for each BMP, to determine the most cost-effective BMP (or combination of BMPs) in the area.
4
1.3 Objectives
The study’s main objective is to identify the most cost-effective suite of BMPs to reduce phosphorus loadings in the study area, based on agricultural, climatic, hydrological and topographical data.
Specifically, we aim to:
1. Simulate effluent runoff in the Ouachita watershed by utilizing the data in the
Mapshed software package.
2. Analyze the data to determine the reduction coefficient of each BMP in the
watershed.
3. Compare this data with cost of implementation data for each BMP to determine
the most cost-effective BMP or combination of BMPs for the study region.
1.4 Overview of Thesis
The thesis is organized as follows: Chapter 1 details the project’s background and outlines the objectives. Chapter 2 provides a literature review, which covers the history of water quality policy in the United Sates and highlights empirical studies of BMP efficiency and the role of optimization and GIS in BMP studies . Chapter 3 contains the data and methods used in this study. The results of the analysis are provided in Chapter 4.
Finally, Chapter 5 summarizes the implications and conclusions drawn from this analysis.
5
Chapter 2. Literature Review
2.1 Water Policy: A brief history
2.1.1 Federal Water Policy The first government regulation establishing any type of limitation on water pollution began with the Refuse Act of 1899. The purpose of this act was not to ensure clean drinking water or preserve the quality of fisheries but to prevent impediments to navigation. This legislation implies that some rivers had become so overfilled with solid waste that it impaired the ability of ships to navigate commonly used routes and waterways. While this is an archaic law with little practical use in present day policy, it serves as a reminder of the humble beginnings of water pollution policies and how badly polluted the waterways of the United States once were (Freeman, 2000).
Presently, the primary legislation governing water pollution in the United States is the Clean Water Act, initially passed in 1948 (Copeland, 2013). This was the first federal regulation to deal directly with more modern instances of water pollution. This legislation enabled the federal government to conduct research dealing with water quality problems and loan funds to local governments for the building of sewage treatment plants.
However, no water quality standards were established and no forms of enforcement were implemented.
The CWA was amended in 1956 and updated in 1965 with the establishment of the Water Quality Act (WQA). The 1956 amendments established an allocation of funds for federal grants, which could be used to cover up to 55% of the construction costs of
6
municipal sewage treatment plants. The amendment also allowed the government to convene meetings between interested parties if serious pollution problems occurred along interstate waterways. These revisions again failed to mandate effluent limits at the individual, watershed, or state level. The Water Quality Act of 1965 sought to remedy this failure by asserting that states must set minimum water quality standards for interstate waterways. To enforce these minimum standards, states would determine the maximum allowable discharge of various pollutants and then distribute permits to major polluters. The basis for permit distribution and the punishments for violating those permits were left to the state’s discretion. While the establishment of water quality standards was an important step in the progress of water policy, the Water Quality Act of
1965 was ultimately a failure due to the costs of monitoring and regulating the waterways as well as a varying dedication among states to water quality control (Freeman, 2000).
The Cuyahoga river fire of 1969 was cause for water policy change and led to a major revision of the CWA in 1972. Water quality issues were also a significant motivation for the establishment of the Environmental Protection Agency (EPA) in 1970
(Fisher-Vanden et al., 2013). The goals outlined in the CWA were: 1) the attainment of fishable and swimmable waters by July 1, 1983 and 2) the elimination of all discharges of pollutants into navigable waterways. The establishment of goals, methods and accountability represented a significant change from earlier federal policy (Freeman,
2000). While these deadlines were not met in many areas of the country and have been postponed through several amendments, their influence on future water policy should not be underestimated (King, 2005).
7
The CWA was revised several times throughout the 1970s and 1980s. The first major revision of the CWA occurred in 1972 as mentioned above. The 1972 revisions substantially increased federal subsidies, established new goals and deadlines for pollution removal and established new regulation and enforcement methods for municipal waste treatment plants. The 1972 revisions also shifted responsibilities for issuing water quality permits to federal authorities. The CWA was subsequently updated in 1977. The
1977 act extended some deadlines established in the 1972 provisions and made clearer delineations between conventional pollutants and toxic water pollutants. The CWA amendments of the 1970s deal largely with PS pollution; however, they do have some minor provisions for NPS. The section dealing with NPS calls for the establishment of area-wide waste treatment management plans (Freeman, 2000).
The 1987 amendments to the WQA are the first federal legislation to seriously address sources of NPS and NPS mitigation. This amendment establishes that states are responsible for addressing NPS problems within their borders. It stipulates that states must identify NPS sources, establish water quality goals and implement management practices to meet these goals in state NPS assessment reports and state management programs. These plans must identify significant sources of NPS and BMPs to diminish these sources. The 1987 amendment also authorizes the EPA to provide grants to assist with the implementation of BMPs approved by the EPA (James, 2011).
The 1987 amendments to the WQA were the last major amendments to the WQA or CWA; however, there have been several important policy programs related to agricultural runoff and NPS since that time. The Rural Clean Water Program (RCWP)
8
was conducted from 1980 to 1990. The program’s stated goals were to: 1) improve water quality in the project area in the most cost-effective manner, 2) assist farmers in reducing
NPS water pollutants in order to meet water quality goals and 3) develop and test programs, policies and procedures for the control of agricultural NPS pollution. The program funded twenty-one test projects across the U.S., which implemented BMPs and monitored their effectiveness. The RCWP helped improve targeting and use of BMPs in the U.S.
Section 303(d) of the CWA requires states to identify lakes, rivers and streams that do not meet current water quality standards and requires municipal and industrial polluters to implement technology-based controls to mitigate the pollution causing these impairments. For each impaired waterway, states are required to establish TMDLs, which set maximum levels of pollution for each water body, which they then submit to the EPA for approval. If states fail to establish satisfactory TMDLs, the EPA is allowed to set a priority list of waterways for each state and establish its own TMDLs. Establishing
TMDLs requires quantitatively assessing both the amount of pollution and the need for pollution reduction in a given waterway as well as establishing the sources of pollution.
TMDLs are applicable to both NPS and PS sources. While these stipulations exist in the initial 1972 CWA, difficulties related to NPS and a lack of EPA funding to pursue these programs detracted from the establishment of TMDLs. After being largely overlooked for more than two decades, several lawsuits by environmental groups led the EPA to propose new rules and guidelines for monitoring and assessing TMDLs in 1997. Congress eventually passed these proposals into law in 2000. Since that time, the EPA has been
9
establishing TMDLs for priority waterways in several states in accordance with the new guidelines (Copeland, 2003).
In order to grasp the true meaning of Section 303(d) of the CWA, it is important to define TMDLs and understand how they are established. The EPA defines a TMDL as a written plan and analysis established to ensure that a waterbody will attain and maintain water quality standards, including consideration of existing pollutant loads and reasonably foreseeable increases in pollutant loads. It is intended to provide an opportunity to compare relative contributions from all sources and consider technical and economic trade-offs between point and nonpoint sources (USEPA, 2012). The stated purpose of establishing each TMDL is to set in motion a series of actions that allocate pollutants in such a way that water quality standards are achieved. Each TMDL outlines maximum allowable pollutant loads to achieve water quality standards for defined critical conditions. Each TMDL must specify the pollutant for which it is established and the amount that may be present to meet water quality standards. The TMDL must also specify the amount the waterbody deviates from the load required to attain water quality standards. The TMDL must take into consideration all point sources, nonpoint sources and consideration for seasonal variations. Each TMDL must allocate pollutant loadings to specific point sources (for example, sewer overflows or abandoned mines) as well as allocations for estimates of nonpoint pollutants. It must also include a margin of safety to account for uncertainty and lack of knowledge as well as considerations for future growth
(USEPA, 2012).
10
The Food Security Act of 1985 (also known as the 1985 Farm Bill) established the Conservation Reserve Program (CRP). The goals of the CRP program were to reduce soil erosion, increase and improve wildlife habitats, protect the nation’s long-term capability to produce food and fiber, provide income support for farmers to curb the production of surplus commodities, protect ground and surface water by reducing runoff and sediment and to help clean lakes, rivers, streams and ponds. The program allows farmers to enter into 10-15 year contracts with the USDA to take highly erodible lands out of production and receive rental payments for returning the land to permanent vegetative cover (Glaser, 2012).
The Environmental Quality Incentive Program (EQIP) is a program created under the 1996 Farm Bill. The initial annual funding for the program was $200 million, but that figure has increased steadily, with annual funding reaching $1.8 billion in the year 2012.
The program was created to assist famers and livestock producers in making environmental and conservation improvements. Under EQIP, landowners establish and implement conservation plans for which they receive cost share or incentive payments.
The goal of this program is to select projects that maximize the benefit of payments made under EQIP. Emphasis is placed on planning to identify current problems and practices capable of addressing these problems (USEPA, 2013).
The 2002 Farm Bill Conservation Provisions (FBCP) provided technical and financial assistance to farmers interested in conservation and improvement of natural resources. The bill introduced or updated several funding and incentive programs for
BMP implementation. The FBCPs include an array of programs that target different
11
BMPs and conservation practices and provide assistance through incentive payment and cost share programs (USEPA, 2003).
In 2008, Congress adopted the Farm Security and Rural Investment Act. This act, alternatively known as the “Farm Bill,” established programs to provide assistance to farmers and ranchers implementing BMPs to reduce NPS pollution, restore wetlands and improve wildlife habitat. The Farm Bill also incentivizes agricultural landholders to participate in several other programs aimed at increasing the utilization of BMPs on their land (LDEQ, 2012).
2.1.2 Louisiana State Water Policy In the state of Louisiana, the primary document addressing water quality is the
Louisiana Water Quality Management Plan (WQMP). The plan was developed in accordance with the policies laid out in the CWA. The WQMP is designed and implemented by the Louisiana Department of Environmental Quality (LDEQ). The stated goal of the WQMP is to ensure the waters of the state meet established water quality standards and thereby maintain all designated uses (LDEQ, 2004).
The CWA mandates that the governor of each state submit an NPS management plan to the EPA. This plan must address: (1) a description of BMPs to be implemented by the state to reduce NPS, (2) a description of management programs utilized to achieve implementation of BMPs, (3) a schedule of milestones to achieve implementation of
BMPs, (4) a certification by the Attorney General of Louisiana that the state water pollution control agency has adequate authority to implement the above policies, (5) a description of federal and state assistance that will be utilized to implement the state’s
12
NPS management program and (6) an identification of other federal financial assistance or development projects the state will review for their effects on the water quality and consistency with the state’s NPS management program. Louisiana received approval for its NPS Management Plan from the EPA on November 21, 2012 (LDEQ, 2012).
The Louisiana NPS Management Plan details ongoing and future water quality projects in the state of Louisiana. The stated goal is to reduce NPS impairments in at least
40 water bodies by October 2016. The report also documents recent water quality improvements undertaken by the state in the years preceding the NPS plan. Previously adopted goals (2005) included reducing the number of waterways on the impaired water bodies list by 25% for three different categories. These categories were primary contact recreation (PCR), secondary contact recreation (SCR) and fish and wildlife propagation
(FWP). This goal was to be achieved by the end of 2012. While the LDEQ was successful at meeting the goals for PCR and SCR, it restored only 8 of the proposed 77 waterways for FWP. FWP waterways are the primary water bodies impaired by nutrient and sediment loadings (LDEQ, 2012).
In addition to detailing past efforts to control NPS pollution, the plan also outlines the state’s efforts to identify and control waterways impaired by NPS pollution. This plan includes an effort to abate known NPS water quality impairments, identify and address new impaired waterways and threatened waterways through the development of
TMDLs and manage and implement NPS programs efficiently and effectively, including financial management and the periodic review and evaluation of NPS management programs using environmental and functional measures of success (LDEQ, 2012).
13
Louisiana also releases an annual NPS report to update interested parties on their goals and achievements for the past year. According to the 2012 report, 23 impaired waterways have been fully restored in the past 12 years. Louisiana also estimates that it mitigated 519 million pounds of nitrogen, 129 million pounds of phosphorus and 89 million pounds of sediment from government-funded projects in 2012. The state also developed 12 implementation plans, identified 20 priority areas for BMP placement and monitored water quality at 14 sites downstream of locations where BMPs had been introduced. While Louisiana has made strides toward achieving water quality goals by identifying impaired watersheds and implementing BMPS, it still has more than 300 impaired waterways and will, in all likelihood, be working to restore its waterways for the next several decades (LDEQ, 2012).
2.2 BMP Efficiency
BMP efficiency has been the focus of hundreds, if not thousands of studies over the past thirty years (Evans et al., 2012). While a comprehensive storm water BMP database exists, and a similar agricultural BMP database was undertaken by the EPA, the agricultural database was placed on hold before its release to the public (Wieland et al.,
2009). Several independent literature reviews and meta-analyses have been conducted to evaluate the pollutant reduction and cost efficiencies of many different agricultural BMPs
(Harmel et al., 2006; Wieland et al., 2009; Merriman et al., 2009; Yagow et al., 2002;
Evans et al., 2012). These analyses draw on numerous sources, from a combination of previous literature reviews (Evans et al., 2012) to a meta-analysis of over 100 site specific BMP studies (Merriman et al., 2009). The remainder of this section will provide
14
a description of the methods used by each of the BMP meta-analyses as well as the information each analysis provides about the BMPs considered in this study.
Wieland et al. (2009) analyzed the efficiency of BMPs to reduce nutrients and sediment specifically tailored for use in the Chesapeake Bay. This study relies heavily upon a literature review conducted by Simpson et al. (2007), which also focused on the
Chesapeake Bay. Simpson et al. (2007) utilize a literature review of current studies as well as an “adaptive management approach” which allows for the application of the best applicable science and the best professional judgment to further adapt the estimated efficiencies derived from the literature review. The Wieland et al. (2009) study summarizes and updates the 2007 study and is referred whenever it reveals pertinent new data. Merriman et al. (2009) compiled a BMP efficiency database with the goal of developing a BMP efficiency tool. They considered a range of sources including some unpublished sources (the study consisted of 18% unpublished sources and 82% published sources). This database reviews a wide range of studies including lab studies, field studies, paired watershed studies and modeling studies. This database averages these sources to find mean BMP efficiencies. Yagow et al. (2002) developed a BMP database of published literature for comparison of BMP efficacies in nutrient load and concentration. This database initially reviewed 596 articles, including only articles that offered primary research and results from field-monitored studies. After considering 596 articles for these criteria, 168 articles were incorporated into the database. After the initial formulation, the database was updated regularly until 2006. Evans et al. (2012) reached efficiency figures for different categories of BMPs after reviewing the work of Yagow et al. (2002), Ritter et al. (2001), Susquehanna River Basin (1998) and U.S. EPA (1990). 15
The goal of Evans et al. (2012) is not to create an in-depth literature review database, but baseline BMP efficiencies for PRedICT, a BMP assessment tool. Unfortunately, these reviews do not contain efficiencies for all BMPs across all pollutants. Whenever a study or efficacy is omitted for a certain BMP, it can be assumed that no figures were available from that particular study. Further complicating comparison is a lack of homogeneity across practice names and a lack of consistency even within a well-defined BMP. All of the aforementioned articles attempt to address this problem by assessing BMP suites instead of individual BMPs. This alone may account for some of the variability in BMP efficiency estimates.
The term “cover crop” refers to the practice of planting crops including grasses, legumes and forbs for seasonal cover and other conservation practices. Cover crops can also reduce erosion from wind and water, increase soil organic matter content, capture and recycle nutrients in the soil profile, suppress weeds and increase biodiversity (NRCS,
2013). Common examples of cover crops include ryegrass, legumes, sorghum, wheat, rapeseed and barley (EPA, 2012). Simpson et al. (2007), utilizing a weighted literature review (75% of literature review coefficients), find that the reduction efficiencies for cover crops were 26% of N loadings for coastal plains and 20% of N loadings for non- coastal plains. Merriman et al. (2009) find that cover crops reduce an average of 66% nitrogen, 67% of the total phosphorus and 70% of the total sediment in the area in which they are implemented. The Virginia Tech agricultural BMP database (Yagow et al., 2002) reports that cover crops give a phosphorus reduction of 48%. Evans et al. (2012) recommend that efficiencies of 25% nitrogen reduction, 36% phosphorus reduction and
25% sediment reduction be used when assessing the effectiveness of cover crops. 16
Conservation crop rotation refers to the practice of growing crops in a planned sequence on the same field. This can improve the soil quality, reduce erosion and manage the balance of nutrients (NRCS, 2013). Merriman et al. (2009) found the average BMP efficiency for conservation cover crops to be 67% for nitrogen, 60% for phosphorus and
72% for sediment. The Virginia Tech BMP database estimates the range for conservation crop rotation 6.8% efficiency for N, 39.9% efficiency for P and 38.1 to 55.4% efficiency for sediment reduction. Evans et al. (2012) recommend BMP efficiencies of 8% for nitrogen, 22% for phosphorus and 30% for sediment.
Conservation tillage refers to the practice of managing the amount, orientation and distribution of crop and other plant residue on the soil surface year round, limiting soil disturbance activities to those necessary to place nutrients, condition residue and plant crops. Tillage can take many forms including conventional till, no-till and mulch tillage. Conservation tillage is broadly defined as tillage that leaves a minimum of 30% of the soil surface covered by crop residue (NRCS, 2013). Harmel et al. (2006) compiled the results of 40 studies analyzing the effectiveness of N and P reduction using different types of conservation tillage practices in different areas of the country. The range for the median total N coefficient is 2%-82% and the range for median total P coefficient is -
12% to 40% depending on the method used. Merriman et al. (2009) find that the average reduction of total P in the application area is 55%, the average reduction in total N is 53% and the average reduction of total sediment is 66%. Simpson et al. (2007) report conservation tillage reduction coefficients of 8% for nitrogen, 22% for phosphorus and
30% for sediment. The Virginia Tech BMP database (Yagow et al., 2002) finds that conservation tillage reduces nitrogen loadings by an average of 68.2% to 87.5%, 17
phosphorus loadings by an average of 18% to 92% and sediment loadings by an average of 22.6% to 67.3% depending upon the technique. Evans et al. (2012) find that conservation tillage reduction coefficients are best estimated at 50% for nitrogen, 38% for phosphorus and 64% for sediment.
A grade stabilization structure is used to control the grade and head cutting in natural or artificial channels. These structures are used to prevent the formation or advancement of gullies. These structures improve water quality by reducing sediment and sediment bound pollutants (NRCS, 2013). The Virginia Tech BMP database (Yagow et al., 2002) find that grade stabilization structures reduce nitrogen by an average of 56.1%, phosphorus by an average of 60.4% and sediment by an average of 82.2% in areas where they are installed.
Nutrient management is a commonly utilized agricultural BMP characterized by a variety of different techniques. These techniques include varying the fertilizer form and rate, varying nutrient application methods and timing, treating soils and manure to reduce the availability and mobility of nutrients and developing a comprehensive farm-wide plan to manage nutrients from all sources (NRCS, 2013). A nutrient management plan is developed to optimize crop yields while minimizing the amount of nutrients leaching from the farm. This is achieved by finding the optimal nutrient balance so that (ideally) nutrients are neither over- nor under-applied (Evans et al., 2012). The Merriman et al.
(2009) meta-analysis found the mean value of nutrient reduction coefficients for a nutrient management plan to be 10% for nitrogen and 48% for phosphorus. The Virginia
Tech BMP database lists the average nitrogen reduction coefficient as 40.7% and average
18
phosphorus reduction coefficients of -19%. Evans et al. (2012) estimate that nutrient efficiency reduction coefficients are 70% for N and 28% for P. Sediment values are not given for nutrient management plans because the plans are aimed at managing applied nutrients and therefore do not reduce sedimentation runoff.
Retirement of agricultural land refers to the practice of returning agricultural land to a state of vegetative or forested cover. This can include conversion of agricultural land to wetlands or forests (Evans et al., 2012). This can also include establishing conservation cover defined as perennial vegetative cover to protect soil and water resources on land retired from agricultural production. In many cases, the land retirement is administered through the Conservation Reserve Program (CRP). This program pays farmers yearly rent
(for a contract period of 10-15 years) to remove environmentally sensitive land from agricultural production and plant species that will improve environmental health and quality (NRCS, 2013). Simpson et al. (2009) find that, in the Chesapeake Bay
Watershed, converting agricultural land to wetlands has an efficiency ranging from 7%-
25% depending on area characteristics, a phosphorus efficiency range of 12%-50% depending on area characteristics and a sediment efficiency of 15% regardless of area characteristics. Merriman et al. (2009) find wetlands to have an average nitrogen reduction rate of 64% and an average phosphorus reduction rate of 72%. Evans et al.
(2012) estimate the coefficients for the retirement of agricultural land to have a nitrogen reduction of 95%, a phosphorus reduction of 95% and a sediment reduction of 95%.
Vegetative buffers (also referred to as riparian buffers, grassed waterways or filter strips) are permanent strips of stiff, dense vegetation established along the general
19
contour of slopes or across concentrated flow areas (NRCS, 2013). Simpson et al. (2007) estimated that riparian forested buffers reduce total N by an average of 38%, total P by an average of 40% and total sediment by an average of 53.3% when considered over a range of land-use types. Merriman et al. (2009) found that filter strips reduce total N by an average of 54%, total P by an average of 57% and total sediment by an average of 56% in the regions in which they are applied. The Virginia Tech BMP database (Yagow et al.,
2002) finds that vegetative buffers have a range of P reduction from 3.6% to 25.5% depending on the type of vegetative buffer. Evans et al. (2012) find that vegetative buffers provide reduction coefficients of 64% for nitrogen, 52% for phosphorus and 58% for sediment.
Fencing refers to the practice of constructing a barrier to facilitate the movements of animals, people or vehicles (NRCS, 2013). For the Chesapeake Bay, Wieland et al.
(2009) estimate the reduction efficiencies to be 25% for total nitrogen, 30% for total phosphorus and 40% for total sediment based on suggestions from previous literature
(Simpson et al., 2007). The Merriman et al. (2009) literature review finds that the mean reduction efficiencies for fencing are 78% for nitrogen, 75% for phosphorus and 83% for sediment. Evans et al. (2012) find that streambank fencing has efficiencies of 56% for nitrogen, 78% for phosphorus and 76% for sediment.
Streambank stabilization structures refer to treatments used to stabilize and protect the banks of streams or constructed channels and shorelines of lakes, reservoirs and estuaries (NRCS, 2013). Merriman et al. (2009) estimate that streambank stabilization structures have a nitrogen efficiency of 78%, a phosphorus efficiency of
20
76% and a sediment efficiency of 83%. Evans et al. (2012) find that streambank stabilization has nutrient and sediment efficiencies of 95%.
The wide range of reduction in BMP coefficients is not provided to suggest that
BMP effectiveness cannot be determined or that these coefficients cannot be estimated for a given watershed. These variations highlight the fact that BMP efficiency is site specific and will depend on soil, topography, crops and vegetative cover, climate, management and maintenance. This variation among analysis and meta-analysis is presented to underscore the need for modeling at the watershed level. As the next section highlights, BMP effectiveness estimates have been combined with local BMP cost data and implemented in several watershed level studies to improve the efficacy of BMPs in the area.
2.3 Optimization and GIS in Determining BMP Cost Effectiveness
Cost-effective NPS reduction in agriculture relies upon the correct selection and placement of BMPs within a given watershed. Factors such as land use, soil variety, topography, hydrology, meteorology and interaction with other BMPs all determine the effectiveness of installed BMPs. Because these factors vary throughout different watersheds, site-specific BMPs are required to reduce NPS runoff in the most efficient manner. To determine which combination of BMPs is best for a given watershed, alternative BMP scenarios must be considered (Veith, 2004).
Several studies have reported the effectiveness of various BMPs at reducing nutrient loads. The amount by which a BMP reduces the nutrient loading is known as the
“BMP reduction factor.” These factors differ among nutrients for any given BMP and 21
vice versa. Studies range in their assessment of the effectiveness of each BMP, largely because the scale of the test area (e.g. plot, field, farm, watershed) and aforementioned environmental factors vary greatly among studies. However, a generally accepted range of the BMP reduction factor can often be determined (Rao, 2009).
The use of GIS-based runoff simulation modeling coupled with an optimization algorithm to estimate the placement of BMPs for nutrient reduction has been explored extensively (Gitau et al., 2004; Veith et al., 2004; Kaini et al. 2012; Alminagorta et al.,
2012). Simulation-based modeling incorporates scientific knowledge to quantify site and
BMP specific response. Optimization allows for variation in spatial factors across a multitude of variables and circumstances. Through the use of optimization algorithms,
BMP interaction as well as a range of site-dependent characteristics can be assessed
(Veith, 2004).
Computer models have long been used to simulate environmental and meteorological occurrences. Arcview’s GIS software provides graphical support for these simulation models and extracts data from digital maps. GIS software also prepares data in the form utilized by most simulation software (Abbaspour et al., 2007). The use of GIS software has become relatively widespread due to the inherent advantages of manipulating spatial data in GIS. The Mapshed simulation software manipulates GIS shape and grid files as well as other non-spatial data to estimate NPS runoff (Evans et al.,
2012).
The term “model” refers to a set of equations or algorithms that are used to simulate a physical system. A multitude of watershed simulation models are available to
22
water quality researchers including: the GWLF model (Haith et al., 1987), the Soil and
Water Assessment Tool (SWAT) (Arnold et al., 1998), the Erosion Productivity Impact
Calculator (EPIC) model (Williams et al., 1984), the Hydrologic Simulation Program
Fortran (HSPF) model used by the Environmental Protection Agency (Donigian et al.,
1984), the Spatially Referenced Regressions on Watershed Attributes (SPARROW) model (Smith et al., 1997), the Soil Moisture Distribution and Routing (SMDR) model
(Zollweg et al., 1996) and the Long-Term Hydrologic Impact Assessment (L-THIA) model (Bhaduri et al., 2000). Models vary in their complexity and assessment capabilities. More complex models require a larger amount of data and make fewer assumptions. It is generally recommended that the simplest model that will sufficiently identify BMP placement be used. However, the model must be capable of quantifying the potential response of the given watershed to site-specific changes. In an EPA-sponsored study, Shoemaker et al. (2005) provide a comprehensive overview of more than 65 watershed modeling programs, providing detailed information about model type, level of complexity and water quality factors assessed by each model. This study found that existing models can simulate the dominant pollutant types and water bodies using available technology.
For this study, we will utilize the Generalized Watershed Loading Function-
Enhanced (GWLF-E). GWLF has been utilized to predict nutrient loads in the Delaware
River (Schneiderman et al., 2002) as well as evaluate BMPs in the San Jaqoiun River
Valley (Cryer et al., 2001) and the Big Cypress Creek Watershed (Santhi et al., 2006).
This model has also been used in watershed analysis in the Hudson River watershed (Lee et al., 2000), the NYC water supply watersheds (Schneiderman et al., 2002) and the 23
Chesapeake Bay watershed (Lee et al., 2000). Evans et al. integrated GWLF with economic models for cost analysis in a 2002 study of watersheds in Pennsylvania.
The use of optimization in combination with an NPS pollutant runoff model has been shown to improve BMP cost-effectiveness. Ancev (2003) determined that simulation-recommended BMP changes reduced phosphorus loadings in the Eucha-
Spavinaw watershed near Tulsa, OK. Srivastava et al. (2002) demonstrated a 56% reduction in pollutant loading and a 109% increase in net profits through the use of a simulation model with an optimization algorithm. Veith et al. (2003) used a GIS model combined with a genetic algorithm to reduce NPS pollutant flows in a 1,014-ha watershed in Virginia.
Several studies have been performed about the cost of reducing pollutants in watersheds throughout the United States. In a study of the Louisiana dairy production region, Hall (2009) estimated the cost of reducing one pound of nitrogen to be $14.60, one pound of phosphorus to be $238.47 and one pound of sediment to be $0.44. The total cost of reduction in the watershed was $37.3 million. This cost occurred by adopting a combination of cover crops, conservation tillage, riparian buffer, critical area planting, nutrient management, vegetative buffer and prescribed grazing. The estimated cost per unit of pollutant with the BMPs currently adopted in the watershed was $70.51 per pound of nitrogen, $819.39 per pound of phosphorus and $1.18 per pound of sediment. The total cost of adoption for the watershed with the current suite of BMPs was $107.7 million.
For two watersheds in Indiana, Arabi et al. (2006) found that combining a watershed-modeling tool with an optimization procedure provided improved cost
24
efficiencies. In the first watershed included in the study, Arabi et al. (2006) determined that the same amount of pollutants could be reduced for a cost of $165,370 as were currently being reduced for a cost $414,690. For the same watershed, this study also determined that for the cost of $414,690, nearly three times the amount of pollutants could be removed than under a targeting scenario with the same cost. For the second watershed included in the study, Arabi et al. (2006) determined that five times the amount of pollutants could be reduced using an optimization model in place of a targeting model for the cost of $60,610.
Maringati et al. (2011) also used an optimization procedure in concert with a watershed-modeling tool to estimate pollution in north central Indiana. Their goal was to achieve maximum pollutant reduction while minimizing cost. They found a range of cost solutions from $25-$275/ha that provided pollutant reductions of 23%-49% for nitrogen,
37%-76% for phosphorus and 45%-83% for sediment.
Alminagorta et al. (2012) use a linear optimization program to determine the most cost-effective BMPs for phosphorus reduction in Echo Reservoir, Utah. For each of the three sub-watersheds included in this study (Chalk Creek, Weber River Below and Weber
River Above Wanship), it is determined that nutrient management is a cost-effective
BMP. Protected grazing land and streambank stabilization are also found to be cost- effective BMPs in Chalk Creek. The total cost of reduction in Chalk Creek is $367,000 for a total phosphorus reduction of 4.4 (metric) tons and a unit cost of reduction of
$83.81/kg. The total cost of reduction in Weber River Below Wanship is $158,000 for a total reduction of 0.9 tons and a unit reduction cost of $167.73/kg. The total cost of
25
reduction for Weber River Above Wanship is $460,000 for a total reduction of 2.7 tons of phosphorus and a unit reduction cost of $167.45/kg.
Gitau et al. (2004) examined the cost of reducing phosphorus loading at the farm level using an optimization procedure coupled with a GA and a watershed-modeling tool.
They found an optimal solution using four BMPs, contour strip cropping, a nutrient management plan, riparian forest buffer and a strip cropping nutrient management plan combination. The target phosphorus reduction set for the optimization procedure was
60% of the 1,471 kg of estimated runoff under a scenario with no BMPs. The total phosphorus reduced in this study was 884 kg for a cost of $1,430. The unit reduction cost for phosphorus under this BMP scenario was $1.62/ kg.
Kaini et al. (2012) use a GA and a watershed modeling tool to estimate costs for
20%, 40% and 60% pollutant load reductions for a sub-watershed in Illinois. For a 20% load reduction scenario, 552,586 kg of nitrogen, 108,524 kg of phosphorus and 20,978 tons of sediment are reduced for a cost of $1.036 million. The unit cost of reduction under this scenario is $1.87/kg for nitrogen, $9.55/kg for phosphorus and $49.39/ton for phosphorus. For a 40% load reduction scenario, 420,760 kg of nitrogen, 81,691 kg of phosphorus and 21,294 tons of sediment are reduced for $1.494 million. The unit cost of reduction under this scenario is $3.55/kg for nitrogen, $18.29 kg for phosphorus and
$70.16 per ton of sediment. Under a 60% reduction scenario, 296,247 kg of nitrogen,
57,865 kg of phosphorus and 20,398 tons of sediment were reduced for a cost of $7.461 million. The unit cost of reduction under this scenario is $25.18/kg of nitrogen,
$128.94/kg of phosphorus and $365.77/ton of sediment.
26
Chapter 3. Data and Methodology
3.1 Data and Study Area
3.1.1 Study Area
Figure 3.1 Louisiana Broiler Production Region
This study investigates runoff from agriculture in the broiler production region of
Louisiana, picture in Figure 3.1. This region, located in north central and northwestern
Louisiana, has a high concentration of broiler production. Poultry production is
Louisiana’s largest animal industry. It contributes $1.5 billion to the state’s economy.
Broiler production makes up a large portion of Louisiana’s poultry production industry.
In 2012, broiler producers produced 912.7 million pounds of broiler meat with a gross 27
farm value of $876.1 million (LSU AgCenter, 2012). Louisiana parishes with a large concentration of broiler production include Bienville, Claiborne, Jackson, Lincoln,
Natchitoches, Ouachita, Red River, Sabine, Union, Vernon, Webster and Winn. These parishes form the region known as the Louisiana Broiler Production Region (LBPR). Due to the structure of the underlying watershed, data on Bossier, Caldwell, De Soto, Grant and Rapides parishes was also included. The tributaries in this region are part of the
Ouachita River Basin, which feeds the Mississippi River. This study area was chosen because it has several watersheds from the Louisiana DEQ’s list of priority watersheds and has not yet been comprehensively studied with respect to BMPs (LDEQ, 2013).
From within the study area, we selected the Ouachita River Basin. This basin is located primarily within the broiler production region and contains several impaired waterways. The main water body in this basin is the Ouachita River. The Ouachita
River’s source is found in the Ouachita Mountains located in west-central Arkansas. The
Ouachita River flows through northern Louisiana and empties into the Tensas and Black
Rivers, which feed the Red River, which in turn feeds the Mississippi River. The
Ouachita River Basin covers 16,100 kilometers of Louisiana. The land in this area is primarily utilized for agriculture and forestry, with the agricultural lands lying in the flat
Mississippi flood plain and the forested lands lying in the hills between the Red River and Ouachita River.
Within the Ouachita River Basin, one watershed from the LDEQ’s list of priority watersheds has been selected for analysis. The focus of this study is the Lower Ouachita
(HUC: 08040207) sub-basin. The Lower Ouachita lies in Caldwell, Catahoula,
28
Concordia, Jackson, La Salle, Morehouse, Ouachita and Union parishes in northern
Louisiana. This sub-basin contains several impaired watersheds that lie within the LBPR.
Within the Lower Ouachita sub-basin we chose to focus on the Cheniere Creek (Figure
3.2) (HUC: 0804020701) and Bayou Desiard (Figure 3.3) (HUC: 0804020702) watersheds. These water bodies are listed as impaired by the EPA and lie entirely within the Louisiana broiler production region (EPA, 2013).
Figure 3.2 Map of Cheneire Creek Watershed
The Cheniere Creek watershed, highlighted in Figure 3.2, is primarily located in
Ouachita Parish with a small portion in Jackson Parish. It covers an area of 38,800 hectares and centers around Cheniere Creek, which flows between the Ouachita River and Cheniere Break Lake. The crop production area in this watershed measures 5,492
29
hectares with a streambank length of 6,700 meters in agricultural land. While Cheniere
Creek does appear on the EPA’s list of impaired watersheds, no TMDL has been determined for this watershed. The EPA has, however, determined that one of the causes of impairment in this watershed is organic enrichment and oxygen depletion. This can be reduced by adopting practices that reduce nutrient and sediment loadings in the waterbody.
Figure 3.3 Map of the Bayou Desiard Watershed
The Bayou Desiard watershed, highlighted in Figure 3.3, is also primarily located in Ouachita Parish with small portions in Jackson and Caldwell parishes. It covers an area of 56,806 hectares and centers around Bayou Desiard and Lake Bartholomew . The crop production area in the watershed measures 10,629 hectares with 42,000 meters of
30
waterfront land. Both the LDEQ and EPA have listed Bayou Desiard as an impaired waterway. Unlike Cheniere Creek, Bayou Desiard has an established TMDL. The TMDL asserts that Bayou Desiard does not meet fish and wildlife standards. The cause of impairment has been listed as low dissolved oxygen levels as well as organic enrichment.
Further studies have shown that all of these loadings are the result of NPS loadings with none of the loadings resulting from PS polluters.
3.1.2 GIS Layers and Cost Data Mapshed, the GIS based watershed modeling tool used to simulate watershed characteristics, requires several data layers to estimate nonpoint source loading and BMP effectiveness on a given watershed. These GIS layers were collected from a number of sources and transformed in order to meet the specifications of the program.
Several shape (or vector) and grid layers are required by Mapshed. The Basin layer shows the boundaries of one or more watersheds where the modeling is being performed. This layer is acquired from the Louisiana water mapping service
(http://sslmaps.tamu.edu/website/srwp/Louisiana/viewer.htm ) and has been clipped so that only the portions of watersheds in Louisiana are assessed. The county layer is a polygon layer, which shows the parish boundaries and is not used to perform any calculations. The land use/cover layer is a grid layer, which uses 16 distinct land use/cover types to help estimate nutrient flows throughout the watershed. These layers were attained from the Louisiana GIS CD ( http://atlas.lsu.edu ). The stream layer contains line features of stream segments for the study area. The stream layer was acquired from the USGS. The surface elevation (topography) layer is a grid layer, which
31
is used to calculate slope-related data used in the model. It has been obtained from the
Louisiana statewide GIS server ( \\gid-store.lsu.edu\gis ). The physiographic province layer contains areas with different hydraulic parameters. These parameters are warm rain erosion rate, cool rain erosion rate and groundwater recession rate. This layer was digitized from a USGS map of physiographic regions throughout Louisiana. The layers listed above were all obtained in the format required by Mapshed. However, several layers were obtained in a format not compatible with Mapshed and were manipulated in order to meet with Mapshed’s formatting requirements.
The animal feeding operation (AFO) layer contains information on the location of farms as well as animal populations by type. Point shape files contain the location of poultry houses and dairy farms. The poultry houses were digitized from a DOQQ file provided as a base map in ArcGIS 10.3. The dairy farm locations were obtained from the
Department of Health and Hospitals. Animal totals were obtained from the LSU
Agricultural Summary’s five-year summary, which provides agricultural data for the years 2006-2010. The summary provides yearly totals, which are then averaged over five years. Animal totals are averaged over a five-year period to minimize the effect of single year market fluctuations, which may drastically alter the number of animals in a parish for an individual year. We selected the range of 2006-2010 because it represents the most current period in which weather data was available for the selected watershed.
The soil layer contains information on various soil properties such as hydraulic group, erodibility factor and water holding capacity. The map of soil type and soil area was obtained from the Louisiana GIS CD. The hydraulic group, erodibility factor and
32
water holding capacity were procured from Louisiana state soil surveys (which contain soil information by parish). Soil areas often contain more than one soil type. For each soil area, the three soil properties listed above are calculated by multiplying the properties of the individual soil types by the percentage of each soil in the soil area. The percentages of soil types in a soil area were also acquired from the Louisiana state soil survey. A soil grid layer, Soil Phosphorus (Soil-P), is used to estimate the phosphorus content of sediment runoff to nearby waterways. This layer was obtained from the soils lab at the
LSU Department of Plant, Environmental and Soil Sciences.
The weather layer contains point layers of the location of each individual watershed. In this study, eight weather stations are included in the model. Each weather station is linked to a table, which contains data on maximum and minimum temperatures as well as precipitation for that weather station, for the longest time period during which data is available. If more than one weather station is within the watershed, the mean daily temperature and precipitation are used. If no weather stations are within a watershed, the means of the two closest weather stations to that watersheds center are used. Data for this layer were obtained from the National Ocean and Atmospheric Administration’s (NOAA) online climatic database ( http://www.ncdc.noaa.gov/cdo-web/search ). Mapshed requires that data be consecutive for every day of the year with no missing values. In cases where
NOAA data contained missing values, estimates were made by averaging the previous and next day’s totals. Temperature data were multiplied by 0.18 and then added to the number 32 [(temp*9/50)+32] to change tenths of degrees centigrade into degrees
Fahrenheit. Precipitation was divided by 254 to change tenths of a millimeter into inches
[prcp/(2.54*100)]. 33
Cost data were obtained from data the NRCS provided on cost share payments made via the Environmental Quality Incentives Program (EQIP) to recipients in
Louisiana in 2013. In many cases, one type of BMP may have more than one relevant cost associated with it. For BMPs with multiple relevant costs, an average of all relevant costs in the area is taken. This is consistent with the GWLF-E model, which takes into account the range of different practices that may fall under a single BMP heading.
Figures obtained from the NRCS were often in acres (with respect to farmland) and linear feet (with respect to streambanks). Where applicable, cost values given in acres were transformed into hectares by multiplying the values by 2.47, and cost values given in linear feet were multiplied by 3.28 to put them in terms of meters.
Grade stabilization structures present a special case because they are costed and applied in terms of individual structures rather than on a per acre basis. To address this and allow grade stabilization to be compared to other BMPs, they are prorated on a per- hectare basis. To achieve this, it is assumed that, on average, four grade stabilization structures will be applied every 100 acres. This is consistent with observed behavior in the area and confirmed by a Louisiana NRCS agent. To calculate the per hectare cost, the average cost of one grade stabilization structure is divided by 25 to spread the cost over the 25 acres that the grade stabilization structure covers. The per-acre cost is then multiplied by 2.47 to put the cost figure in terms of hectares. Costs for all of the BMPs investigated in this study can be found in Table 3.1.
34
Table 3.1 BMP cost per land unit
BMP $/ hectare $/ meter
Cover Crop $182.32 -
Conservation Tillage $71.93 -
Conservation Crop Rotation $26.82 -
Grade Stabilization Structure $363.65 -
Nutrient Management $41.57 -
Agland Retirement $152.98 -
Vegetative Buffer - $18.88
Fencing - $5.62
Streambank Stabilization - $86.38
3.2 Watershed Modeling
Nonpoint source pollution from agriculture is recognized as the leading cause of water impairment in the United States. Due to its diffuse nature, NPS pollution sources are difficult to detect and costly to monitor. BMPs have been devised, through years of development, to mitigate NPS pollution; however, correct spatial placement of BMPs is essential for pollution mitigation. Moreover, determining the most cost-effective combination of BMPs to reach predetermined daily loadings is essential when dealing with agricultural producers and public policy makers on ever-decreasing budgets.
35
To aid in determining the spatial placement and most cost-effective combination of BMPs to meet pollution reduction targets, watershed simulation models have been developed. These models range in their complexity, but all of them require broad temporal and spatial data that must be assembled, analyzed and interpreted. Computer modeling has seen significant improvements in the past several decades, with increases in computing technology and the development of GIS software. These models are now recognized as essential tools for NPS pollution mitigation.
The Pennsylvania Department of Environmental Protection has developed one such watershed model in concert with Pennsylvania State University to enable NPS simulation in Pennsylvania. Though originally developed for Pennsylvania, the model was designed to be adapted to other states and geographic regions. This GIS based modeling program is known as the Generalized Watershed Loading Function-Enhanced
(GWLF-E) and is an updated version of early AVGWLF and GWLF models. AVGWLF has been updated and renamed Mapshed. In addition to updating AVGWLF, the latest version (Mapshed) runs in an open source GIS program named Map Windows. Older versions of AVGWLF utilized ArcView’s GIS software, which is expensive. For this project, we used Arcview GIS 10.3 to format the data, as it allows more flexibility than
Map Windows and was available via Louisiana State University’s GIS laboratory. The original program, developed for TMDL projects in Pennsylvania, has been adapted for watersheds in the LBPR.
The procedure used in the GWLF-E phase of this research are as follows: (1)
Identify the parishes in the production region which overlay the related watershed, (2)
36
gather environmental and demographic data (weather, soil type, AFOs, etc.) for the watershed, (3) reformat and input compiled data into GWLF-E computer module using
Arcview GIS, (4) identify impaired watersheds in the LBPR and (5) simulate nutrient and sediment runoff at different levels of BMP adoption throughout the selected watersheds.
GWLF-E is used in place of costly onsite monitoring to simulate phosphorus, nitrogen and sediment runoff. Additionally, GWLF-E contains algorithms to simulate pathogen loadings. GWLF-E is a distributed/ lumped parameter watershed model, meaning that it is distributed in surface loading, considering various land use cover scenarios but a lumped parameter model in sub-surface loading. The model is continuous with respect to weather, utilizing daily inputs. Erosion and sediment yield calculations are estimated on a monthly basis and combined with transport capacity, based on watershed size and daily runoff, to determine sediment loadings (Evans et al., 2012). Dissolved phosphorus and nitrogen coefficients are applied to surface runoff to determine surface nutrient losses. Subsurface losses are calculated by using phosphorus and nitrogen coefficients for shallow groundwater. These monthly loadings are then averaged into yearly loadings, which are utilized to determine average loadings over the entire 10-year period.
This analysis focuses on two watersheds in the Lower Ouachita sub-basin, both of which lie in the LBPR. The EPA lists both of these watersheds as impaired. The data layers for each watershed are analyzed individually for spatial accuracy. Mapshed utilizes these data layers to create input files for use in the GWLF-E model over a given time period and growing season. For this study, the years 2001-2010 were selected for the
37
Chenerie Creek watershed and the years 2003-2012 were selected for the Bayou Desiard watershed. These are the most current ten-year periods for which all necessary data are available for each watershed. The crop-growing season is selected as April through
October, which is consistent with the crop-growing season in Louisiana. GWLF-E takes these input files and simulates the runoff of sediment and nutrients (nitrogen and phosphorus) for a watershed using the process detailed above.
Coefficients for both wet and dry years are calculated to estimate the effect that drought years and years with surplus rainfall have on runoff and BMP effectiveness.
Drought and surplus rainfall are determined by summing the total amount of rainfall in each watershed for a given year. The total rainfall for each year is then compared to total rainfall for all other years in the study period. For the Chenerie Creek watershed, 2004 was the year with the highest rainfall with 68.37 inches of rainfall and 2010 was the year with the lowest rainfall with 36.11 inches of rainfall. For the Bayou Desiard watershed,
2004 was the year with the highest rainfall with 70.13 inches of rainfall and 2005 was the year with the lowest rainfall with 33.25 inches of rainfall.
To estimate BMP effectiveness, Mapshed’s BMP land coverage scenario editor is utilized. This scenario editor allows the manipulation of the area of agricultural land or streambank length (depending on the nature of the BMP) that is designated to a specific
BMP. Land coverage is manipulated as a percentage of existing farmland, which is calculated by Mapshed based on input data. Streambank based BMPs are manipulated based on portions of agricultural stream length. This parameter can be manipulated by the tenth of a kilometer; however, for consistency, agricultural stream length application is
38
manipulated as a percentage of total agricultural stream length. The phosphorus, nitrogen and sediment reduction parameters for each BMP can be manipulated using Mapshed’s rural BMP efficiency editor. In addition, estimates were obtained for a 10% increase in each load reduction coefficient and a 10% decrease for each load reduction coefficient for each BMP, thus creating a range of reduction. One hundred and fifty simulations are run for each BMP in each watershed. This allows each BMP to be simulated at 2% steps in land or stream coverage scenarios, for normal, increased and decreased BMP reduction parameters. Each simulation in Mapshed yields an output summary in .csv (spreadsheet) format. These spreadsheet summaries contain, among other statistics, average monthly loading estimates of phosphorus, nitrogen and sediment. Output spreadsheets are reformatted using an R program that results in a summary spreadsheet for each BMP with phosphorus, nitrogen and sediment loadings at different levels of BMP adoption. A separate summary spreadsheet is created for each BMP at 10% increased reduction rates and 10% decreased reduction rates.
3.3 BMP Reduction Coefficients and Optimization Programming
BMP effectiveness is determined by a “BMP reduction coefficient.” These coefficients are representative of the amount of nutrient or sediment reduction provided by a one unit (hectare for field based and meter for stream-based BMPs) increase in BMP adoption. These coefficients are used to quantify BMP effectiveness in the optimization program.
To obtain effectiveness coefficients, regression analysis is performed on the simulation output. Each simulation output is subtracted from a baseline simulation with
39
no BMP coverage to obtain the amount of nutrient reduction at each level of adoption for each BMP. The amount of total agricultural land or streambank length is calculated at 2% intervals, creating a list of corresponding land coverage for each BMP. The amount of nutrient reduction for each level of adoption is then regressed on the amount of land associated with its coverage level. This yields the nutrient reduction coefficient, which indicates how many kilograms (or metric tons for sediment) are reduced per unit of land.
An optimization model is then utilized to determine the ideal land coverage, at the least cost, for different levels of pollutant reduction. The linear programming procedure is detailed below in Figure 3.4. The goal of this optimization program is to minimize cost while maximizing pollutant reduction. To achieve this, constraints are placed on scarce resources as well as minimum requirements for nutrient reduction rates. Phosphorus was chosen as the primary nutrient for reduction because it is recognized as the primary chemical contributing to water pollution, eutrophication and hypoxia in the Gulf of
Mexico. Nitrogen and sediment reduction were also targeted as secondary goals. In each watershed (as well as in wet and dry years), phosphorus reductions of 10%, 15%, 20% and 30% (where feasible) were analyzed. Maximum feasible reductions (determined as a percentage of total phosphorus reduction) were also considered across both watersheds.
Additional constraints were placed on maximum land usage for each BMP.
Agricultural land (Agland) based BMPs are restricted to agricultural land in the watershed, and streambank BMPs are restricted to the total stream embankments in agricultural areas (these figures are taken from GWLF-E, which derives them from the
40